There’s no entry for X in the A-Z. If there were, one of the few candidates would be X-bar theory, which is in fact a (1981) refinement of Chomsky’s theory of language known as Transformational-generative (TG) grammar. (For what it’s worth, X-bar theory argues that all phrases – whether noun phrases, verb phrases etc – have the same structure, and that this structure is a linguistic universal, i.e. it’s common to English, Japanese, probably even Klingon).
Noam Chomsky
As I say, there’s no entry for X-bar theory. There’s no entry for TG grammar in the A-Z, either. Nor for its predecessor, generative grammar. Nor for Government and Binding theory, nor the Principles and Parameters theory, nor the Minimalist program. In fact, there’s no mention of Chomsky or any of his theories in the entry on Grammar at all.
This might strike some readers as odd, even perverse. At best, negligent. After all, the study of TG grammar (or any of its offshoots) is a key component of any self-respecting linguistics course on any MA TESOL program in the US. It is often the only theory of grammar that is studied. In fact, in many of the standard texts, such as Fromkin et al. (2007) An Introduction to Language, or the Ohio State University Language Files (ed. Stewart and Vaillette, 2001) it’s not even called TG Grammar, nor ascribed to Chomsky by name. It’s simply the grammar that is. The one and only. And, just in case you don’t know which one I’m talking about, it’s the one that involves the endless “tree-diagramming” of (invariably invented) sentences, like The child found a puppy and Where has Pete put his bone? (both from Fromkin et al.).
Simple tree diagram
So why did I ignore it?
Well, to be perfectly honest, I’m not sure that I really understand it. I can get my head around basic Phrase-structure grammar, even X-bar theory, and just about understand what Theta-theory is on about. But, as much as I want to get to grips with the Minimalist program (not least because it seems to be arguing a central role for lexis in determining syntactic structure), I’m struggling. In the end, all those upside-down trees leave me cross-eyed.
But there are more cogent reasons – both linguistic and pedagogical – for treating TG grammar cautiously, it seems to me. On linguistic grounds the theory seems flawed since it is based entirely on invented sentences in their written form. Try to apply the descriptive framework to spoken data – e.g. authentic utterances like: But the spa, you might want to use it, you know – and it just doesn’t fit. By the same token, TG grammarians will reject forms as being ungrammatical even when they are commonly attested (one of the texts I consulted disallows the sentence John bought what? for example). Chomsky’s dogged insistence on making the “well-formed” sentence the centrepiece of his theory of language seems to undermine the whole enterprise – this, along with the misguided notion that all sentences are generated from the word up, and are hence all entirely original. (Chomsky’s acolyte, Stephen Pinker, woefully betrays his ignorance of developments in corpus linguistics by claiming – in The Language Instinct – that “virtually every sentence that a person utters or understands is a brand-new combination of words, appearing for the first time in the history of the universe” (1994, p. 22) Compare this to the corpus linguist, John Sinclair’s (1991) claim that “by far the majority of text is made of the occurrence of common words in common patterns”).
The blinkered disavowal of the validity of performance data is, of course, a side-effect of their (Chomsky’s, Pinker’s etc) mentalist agenda, which is to demonstrate both the universality and innateness of their grammar. This means cherry-picking your examples (and re-configuring your theory fairly regularly so as to accommodate new, potentially disruptive, evidence), while consigning anything that doesn’t fit to the damaged goods bin – the one labelled “performance”.
But, more importantly for me, is the lack of pedagogical applicability. For a start, the idealised nature of Chomskyan grammar seems to bear only an accidental relationship to the way language is actually stored and generated in the brain. Chomsky (to his credit) was quick to acknowledge this. Way back in 1965, he wrote: “When we say that a sentence has a certain derivation with respect to a particular generative grammar, we say nothing about how the speaker or hearer might proceed, in some practical or efficient way, to construct such a derivation.” (Aspects of the Theory of Syntax. Cambridge, Mass.: MIT Press, p. 9.) Quite right: the tree diagram describes the structure “after the event”, but it ignores the fact that (as David Brazil put it) “discourse [is] something that is now-happening, bit by bit, in time, with the language being assembled as the speaker goes along” (1995, p. 37). If the TG representation of grammar has little or no psychological foundation, it would seem to be fairly useless for teaching purposes. I take heart, therefore, from a statement made by Bernard Spolsky, after a think-tank on the applications of Chomskyan grammar, held in the late sixties. He concluded:
Linguistics and its hyphenated fields have a great deal to offer to language teachers, but the fullest benefit can only come when their implications are integrated and formed into a sound theory of language pedagogy. Because linguistics is only indirectly applicable to language teaching, changes in linguistic theory or arguments amongst linguists should not disturb language teachers.
(Spolsky, B. (1970) Linguistics and language pedagogy – applications or implications? [emphasis added])
So, why does the teaching of TG grammar (including X-bar theory) persist in the US academic context? And was I wrong to ignore it in the A-Z?
An A-Z of ELT 
Thanks for a fascinating post, Scott, and for presenting some really complex terms and concepts in quite approachable terms!
In answer to your questions…
I don’t know why TG Grammar and X-bar theory persists in US MA TESOL programs, but then again, there are a range of things that I find tedious and unnecessary in MA TESOL programs (which in general I’ve found a little too heavy handed with the “Linguistics” and rather lacking when it comes to “Applied” – the designers of such courses obviously don’t pay all that much attention to Spolsky’s recommendation), but that’s a personal observation!
Were you wrong to leave it out of A-Z of ELT? Yes and no. Yes, because I think it’s good to have such a prevalent theory explained with your observations about what could be wrong with it. No, because I very agree with Spolsky!
🙂
~ Jason
Sorry for that quick-typing-interrupted-by-1-yr-old-daughter at the end: I meant to say: No, because I very much agree with Spolsky!
“I don’t know why TG Grammar and X-bar theory persists in US MA TESOL programs…”
Nor do I, Jason, although there are some ‘conspiracy theorists’ who believe that it’s a plot on the part of the “nativists” to drown out the noise of those who are arguing for experiential, social and cultural factors as being the main drivers of language acquisition. Geoffrey Sampson, in ‘The Language Instinct Debate’ (1997, 2005) argues that, by 1975, “discussion had come to focus on conceptual issues of nature and nurture, rather than on formal properties of grammatical rules.” He goes on: “Perhaps the most important point keeping the nativist hegemony in being is that, to be blunt, there are more jobs in nativism… If people believe that we inherit a rich structure of mental machinery which detrmines many details of intellectual life, then there are careers to be made from teasing out the exact nature of different areas of this innate structure. Many academics earn their livings this way…”
Food for thought?
I saw this in action in Korea when I worked as a ‘visiting professor’ in a department for English Education at one of the top universities there. The local academics in that (and in other departments of English education around the country) specialise in very ‘deep’ linguistic analyses which are incredibly far removed from the realities and needs of actual high school English teachers – the people taking degrees for teaching qualifications at these institutions – but such academics are held in ridiculously high esteem and make all the program decisions. I saw some MA and PhD thesis applications from our students, under the guidance of such academics, and it was very much promoting ‘more of the same.’ Nothing to do with things like affective factors in learning, managing mixed ability and/or very large classes, or assessment (in my opinion, representative of the most relevant and burning issues at the heart of progressive language education in an over-crowded public system), but things like TG Grammar, comparative syntax, incredibly complex semantics… There were some exceptions, driven in my own department by an amazing professor from the UK who has been in Korea for donkey’s years, but it was very clear the preferred and looked-up-to foci were almost purely academic and linguistic in orientation.
Of course, such purely linguistic foci also make for more and ‘tidier’ research papers, the number of which is the determining factor in that context in terms of who earns the most academic prestige and hence positions of influence (not to mention money!). I’m not sure I’d call it an overt conspiracy, but it is a circular self-feeding process which keeps very abstract linguistic thinking at the top of the language education foodchain.
I am only talking about Korea here, but I’m sure there is some resonance with other contexts as well. Personally, I focused on very practical issues with my learner-teachers, and the majority told me they liked my lectures/coursework more (than the academic ones) and felt it was more relevant and helpful for their future roles in public school classrooms. I was happy enough with that, but then again, I had no interest in getting into or competing in the academic (highly linguistics-only) research sphere ‘above me’ in the ELT foodchain.
‘Nuff said! On with the TG Grammar and X-box discussion!
Cheers,
~ Jason
Good topic, post and analysis, Scott, thanks!
You made me recall endless hours of fun in the lecture hall at Reading, trying to understand what Mike Garman was on about and observing the Chinese students, who took to it like ducks to water, twirl him around their fingers in deep and incomprehensible discussions!
We also had a lot of fun using some of the terms in a nonsense sort of way, like creating a pop group called “The Archiphoneme and the Rel Dels”!!! Such nonsense is common on M.A. courses, a time for processing maybe too much with too little time to assimilate it all.
Since then, I have come to rethink and revise my views on this.
Despite your absolutely valid comments regarding the the type of sentences analysed in this model and the difficulty of applying it to naturally occuring, unstructured talk, I find beauty and elegance in the analysis.
Perhaps it appeals to my way of thinking, and there are other people like me out there, who actually, for whatever perverse reason (!), like this type of analysis. I remember asking our teacher of Turkish to give me a tree structure of the sentences I could not make sense of syntactically, and when she did, and I am glad she was able to do it, things fell into place.
So my reaction would be to, yes, do please include it in your rewrite!
There is another reason why I have come to see more importance in the TG model and I am sure it is not a revelation to those who study it – it’s just that I don’t have any relevant literature to quote right this minute to support my point which is this:
If you are looking at understanding and connecting deep structure with propositional meaning, which, in turn, helps in understanding locutarionary only, meaning, then I think you do need to understand TG.
Granted, you need to move beyond the locutionary, but you also need to begin somewhere.
“I find beauty and elegance in the analysis”.
I agree, Marisa – there is a great deal of satisfaction to be gained – at an intellectual and even aesthetic level – out of unpackling the syntax of a complex sentence. But “parsing” sentences can be done just as well using traditional categories, without having to get into the kind of “deep structure” analyis on which transformational-generative grammar depends in order accommodate all the possible transformations in all existing languages.
Moreover, because traditional sentence analysis includes the functional categories (subject, object, adverbial etc) that the Chomskyan analysis deliberately eliminates, the resultant analysis reveals something about the meaning of the sentence – not just its form.
La grammaire générative transformationnelle was such fun to study ! During the student years, the logical minded students could get lots of points with it, and as it was coupled with translation work, it sort of compensated if you needed extra points. But in my practice now in the classroom, I’m not sure GGT helped me. Lack of context, isolated sentences desembody the language. In your “7 ways to look at grammar” speech you suggested to make sure the conditions are there for the “emergent language” to emerge, and then take some time to raise counciousness about the language. In the process of doing this, students inevitably try to find a “pattern”, something they can grasp hold of, and sometimes they can’t have anything very defnite. I think it’s my job to try and find the pattern beforehand. There are areas in French where I can find no solid pattern at all even if looking through the literature. GGT can’t help with this.
“…students inevitably try to find a “pattern”, something they can grasp hold of, and sometimes they can’t have anything very defnite. I think it’s my job to try and find the pattern beforehand.”
I totally agree that a large part of learning a language is identifying, recognising and internalising its patterns – if we know anything about learning styles, the capacity to spot patterns seems to be a style worth cultivating! The teacher can certainly help here – but in the end I think that it’s the patterns that the learners themselves discover that will be both more memorable and more generalisable. Because TG grammar is concerned with patterns at a deeper level of abstraction than the surface phenomena that (arguably) are the raw material of language acquisition (at least, according to the emergentist/usage-based school) it is doubtful whether it is of much use to learners. In this sense, language learning is largely inductive. “The real stuff of language acquisition is the slow acquisition of form-function mappings and the regularities therein. This skill, like others, takes tens of thousands of hours of practice, practice that cannot be substituted for by provision of a few declarative rules” (N. Ellis, 2002, p. 175).
Connectionist models of language processing show that it is possible to abstract grammatical information from the surface features of text: “The networks move from processing frequency relations between mere surface regularities to representing something more abstract, without this being built in as a prespecified syntactic or other linguistic contraint” (N. Ellis ibid.) In other words, the input yields its grammar without there being the need to hypothesise any “hard-wiring”.
Hello, Mr Thornbury
Thank you for the post.
I wonder if there’s V for (Verb) Valency in your alphabet)) That was even more fun than with TG at the University.
The problem is that my ESL students can’t appreciate my deep knowledge of English linguistics. That is more for our mental satisfaction, isn’t it?
They are proud of their teacher being such an expert on ‘all-that-incomprehensible-theory’, I do hope. But my profound understanding of sentence structure, theme-rheme analysis and other God-only-knows theories just can’t make my starters use –s in the third person singular or my intermediate students stop using ‘will’ after ‘if’ in 1st conditionals.
Am I alone in that, I wonder?
Olga (Moscow)
Although, Mr Thornbury, I must say I can’t disagree with your pedagogical reasons for “treating TG grammar cautiously”, as you so nicely put it, I believe your understanding of Chomsky’s linguistic program is somewhat misled. The fact that when it started data weren’t that important can be understood by the overimportance data had for linguistic analysis previously. Chomsky adopted a popperian view of science, that is, let’s assume swans are white and wait for somebody out there to prove us wrong, which is a quite different standpoint from let’s try and see whether each and every swan is indeed white before we get to say anything at all.
Transformational grammar was built around the notion of rule, and although rules such as AFFIX HOPPING might sound a bit naive today, the idea of a category like AUX was pretty original, and in any case made it possible to explain away the syntactic presence of an apparent dummy element like the auxiliary DO in the English language, something that even the great Jespersen wasn’t able to do.
The Principles and Parameters Theory took full advantage of more than twenty years of research on specific transformational grammars. Now the amount of data was large enough for linguistics working within the program to built up a grammar based on “principles”, and no longer on “rules”. And the notion of “parameters” (at first a sort of side-effect to the idea of “principle”, so much so they’ve been referred to as “open principles”) opened up a whole new world of research for applied linguists interested in investigating whether GU would be still operating in second language acquisition (see L. White’s work for instance).
It is possible that the minimalist program has shifted away from data, which is a pity, but then I can’t say much about it for I’ve stopped studying it, and shouldn’t like to venture saying anything about it.
Thanks, Conrado, for your comment. And I agree that TG grammar provides a more elegant description of some aspects of English syntax than most conventional grammars do. Having an AUX category makes the transformation involved in question formation easier to describe. But that’s all it is – descriptive. I’m not sure that it’s explanatory. That is to say, I don’t think the AUX category is part of our grammatical hard-wiring; rather it is a feature that has simply evolved over time as a cultural artefact, much the way that the future marker going to evolved out of a metaphor. After all, the do auxiliary in questions and negatives is a relatively recent phenemenon, and wasn’t firmly in place until the late 17th century. Previously, questions were formed be inverting the main verb and the subject, as in French. Thus, Shakespeare: “What think you on’t?” Historical linguists, who have traced the development of auxiliary do, show that its frequency, relative to the ‘main verb inversion’ pattern, followed a kind of S-shaped evolutionary curve (where there is a period of slow change followed by a period of rapid change) that also characterises the way – in Darwinian terms – one species replaces another. That is to say, the AUX category – for at least three centuries – was in a state of flux. It simply did not exist as part of an innate and invariable mental structure.
This evolving and emergent property of grammar is not recognised in static formalist grammars like TG grammar, which tend to imply that the structures they so elegantly diagrammitise are set in stone. As Larsen-Freeman and Cameron comment, “In order to address their goal, many lingustis have sought to represent the language system in idealized elegance, often stripped of the disorderliness of what had been called the noisy ‘remainder’ … To do so, they have had to make certain concessions that do not cohere well with the demands of applied linguistics”. One of these has been to “freeze” language by removing the dimension of time. “By making synchronic language the object of investigation, the messiness that comes from variation-induced language change is lessened”. (Complex Systems and Applied Linguistics, OUP, 2008).
Thus, TG might serve to capture a snapshot of the language as it is, in one place at one time. But it says nothing about (a) how it evolves over time (b) how it varies across contexts and (c) – most importantly for our purposes – how it emerges in the individual, either as a first or as an additional language.
You’re quite right, Scott, about TG. The AUX category is indeed descriptive. All in all the whole of TG is rather descriptive. But the category evolved to INFL.
This theoretical evolution, if I may say so, was due, first, to the attempts of the French linguist Nilocas Ruwet, in his “Introduction à la Grammaire Générative” (1967). Ruwet believed what was at work in the English grammar could also be appropriate to the French one. Some Brazilian linguists of the early 70s tried the same analysis for Portuguese. Differently from what Ruwet thought was right for French, that is, that its AUX category worked just as it did in English, it wasn’t long until the Brazilians understood that for Portuguese something else had to be conceived. But it was Edmond 1978 (“The Verbal Complex V’ – V in French”, in Linguistic Inquiry, 9, 151-175) who first saw it through: for him, differently from what happened in English, in French AUX didn’t move down onto the verb; it was the verb that moved up onto AUX. We were then left with the question of why in some languages (as apparently was the case with English) it was AUX that moved, whereas in others (French, Portuguese) it was the verb that moved.
The linguist who first proposed an answer to that question was Jean-Yves Pollock, in his seminal article “Verb Movement, Universal Grammar, and the Structure of IP”, in Linguistic Inquiry, 20, 365-424 (1989). For Pollock in all boiled down to morphology. To understand it in a simple way, let’s consider the Latin sentences “Magistra discipulum amat” and “Magistram discipulus amat” (“The teacher loves the student” and “The student loves the teacher”, respectively). Notice that, differently from what happens in English (and in French or Portuguese, for that matter), in Latin the position of the elements is irrelevant. What really matters is the ending (the “desinence”) of the elements. “Magister” and “discipulus”, irrespective of their positions in the sentence, will always be the subject, whereas “magistram” and “discipulum” will always be the object. From this we can gather that movement and morphology seem to be connected in some very interesting way. Now, the fact the verb movement was lost in English can be attributed to its current poor morphology. In Shakespeare’s time a sentence like “What think you on’t?” (to use your example) lived alongside sentences like “whose sore task DOES NOT DIVIDE the Sunday from the week” (Hamlet, I, i, 75-6), in which the presence of DO in INFL can be explained by assuming there was no verb movement here. In the case of “What think you on’t?”, the verb moved. Thus, it is highly probable that in Shakespeare’s time English grammar was going through a stage of hybridness, an idea not too far from possibility since natural processes such as language change are never an over-night process, as you’ve acknowledged yourself above.
But what seems to be relevant here is that, from a purely descriptive device in TG, the analysis of AUX/INFL developed into something whose syntactic comparative powers were now in a position to reach explanatory significance.
Thus, Scott, although I totally agree with you on the existence of a bunch of bullshit in MA programs to do with sheer linguistics, I don’t think we should leave out of our considerations certan linguistic insights when dealing with second language acquisition. At least, Lydia White’s “Universal Grammar and Second Language Acquisition”, and studies referenced there (as more updated ones) could very well have a place in your A to Z review, be it only historical reasons.
Nor the love for Generative Grammar nor the hatred of it should never blind us to the seriousness of studies carried out in the field. Come to think of it, no one so far has got to something around 70% of certainty of how languages (second or first) are actually leant.
Abraço,
Conrado
Thanks Conrado.
For those in the dark, INFL stands for inflection and is a key component in Chomskyan analysis, being to the sentence what the noun is to the noun phrase, i.e. the notional head, as well as being a separate element from the actual verb phrase (VP). To accommodate this component, the notion of Deep Structure has to be invoked, because INFL is not always realised by an independent feature in the surface structure. Instead (in English at least) it is conflated with the verb. She works combines both INFL and VP in works. This is neat when dealing with the auxiliary, e.g. she does work, but less neat otherwise. However you deal with it “the point is that there is mismatch between phonoloy and meaning, which has to be encoded somewhere in the mapping among the levels of structure. If this mismatch is eliminated at one point in the system, it pops up elsewhere. Much dispute in modern syntax has been over these sorts of mismatch and how to deal with them”.
The quote comes from Jackendoff, Foundations of Language, 2002, which I’m currently browsing. Jackendoff is a Chomskyan, in many ways, but he does attempt to resolve some of these paradoxes in his book, as well as creating a model of language processing which marries competence and performance – something Chomsky wouldn’t be interested in. That is to say, Jackendoff attempts to map on to the concept of Universal Grammar findings in psycholinguistics regarding language processing and memory, and even goes so far to show how the UG may have evolved.
(By the way, reading this book I realise I have misdated X-bar Theory – it goes back to 1970)
Indeed, Scott, it goes back to Chomsky’s 1970 article “Remarks on Nominalisation”. Jackendoff wrote a very nice book about it, “X’-Syntax: A Study of Phrase Structure”, published in 1977 by MIT Press. But I believe the most quoted Jackendoff’s work of the period is “Semantic Interpretation in Generative Grammar”, published in 1972 by MIT Press. His proposal for adverbs positioning is very interesting and very much referred to in the literature. The last Jackendoff’s work I read was “Simpler Syntax” (written with Peter W. Culicover), published in 2005 by Oxford University Press.
Languages are sounds that have meanings, right? A grammar of a natural language would then be a system that maps sounds to meanings. Mainstream generative grammar (MGG), as Jackendoff and Culicover name it, has struggled to show this mapping relationship between sound and meaning is done through a computational system which is syntactical in nature. Although this system has been made more and more general through time, subtituting, for instance, a system of principles for another based on rules, it still very much depends on highly abstract and “technological” mechanisms. The idea of the authors of “Simpler Syntax” was thus to reduce syntax to “the minimum structure necessary to mediate between phonology and meaning”.
The book is predominantly critical, since it deals with decades of tradition behind MGG. The form of the argument used by the authors is, I quote, “given some phenomenon that has provided putative evidence for elaborate syntactic structure, there nevertheless exist numerous examples which demonstrably involve semantic or pragmatic factors, and in which such factors are either impossible to code uniformly into a reasonable syntactic level or impossible to convert into surface structure by suitably general syntactic derivation. Generality thus suggests that, given a suitable account of the syntax-semantics interface, all cases of the phenomenon in question are accounted for in terms of the relevant properties of semantic/pragmatics; hence no complications are necessary in syntax.”
Another quotation worth making is the one in which the authors present what they call their “Toolkit Hypothesis”: “The language faculty, developed over evolutionary time, provides human communities with a toolkit of possibilities for cobbling together languages over historical time. Each language, in turn, ‘chooses’ a different selection and customization of these tools to construct a mapping between sound and meaning”.
If it is indeed so, learning a second language is in fact a very complicated sort of undertaking, for we need to learn to use different tools, or else use them in a totally different way.
I must confess I often feel a bit tired, if not lazy, of studying “hardcore” linguistics, but I still believe it can be interesting to understand how second languages are learnt.
Cheers!
Conrado
Speaking as a native speaker studying for an ESL pedagogical degree in Poland who has failed this subject and needs to re-take it in September, I love you!
I’m taking my first linguistics class ever, and it’s not even about my native language, it’s about Japanese. I thought maybe this was the reason I couldn’t quite wrap my head around X bar, since of course the theory was written with English in mind (even though I am at a high level of Japanese as a second language). All I can say is, sitting in class listening to the lecture, I kept thinking that the theory was trying to bend human language into a shape that fit the theory, not bend the theory to fit human language. They were just trying too hard to make it fit. And maybe that’s why I just don’t get it